The volatiles of pathogenic and nonpathogenic mycobacteria and related bacteria

Volatiles released by pathogenic and nonpathogenic mycobacteria, as well as by mycobacteria-related Nocardia spp., were analyzed. Bacteria were cultivated on solid and in liquid media, and headspace samples were collected at various times during the bacterial lifecycle to elucidate the conditions giving optimal volatile emission. Emitted volatiles were collected by using closed-loop stripping analysis (CLSA) and were analyzed by gas-chromatography–mass-spectrometry. A wide range of compounds was produced, although the absolute amount was small. Nevertheless, characteristic bouquets of compounds could be identified. Predominantly aromatic compounds and fatty-acid derivatives were released by pathogenic/nonpathogenic mycobacteria, while the two Nocardia spp. (N. asteroides and N. africana) emitted the sesquiterpene aciphyllene. Pathogenic Mycobacterium tuberculosis strains grown on agar plates produced a distinct bouquet with different volatiles, while liquid cultures produce less compounds but sometimes an earlier onset of volatile production because of their steeper growth curves under this conditions. This behavior differentiates M. tuberculosis from other mycobacteria, which generally produced fewer compounds in seemingly lower amounts. Knowledge of the production of volatiles by M. tuberculosis can facilitate the rational design of alternative and faster diagnostic measures for tuberculosis.


Introduction
Tuberculosis (TB) remains one of the most threatening diseases on earth. In 2008, up to 2 million people died as a result of TB infection [1]. The causative agent, M. tuberculosis, has a highly flexible physiology and metabolism that allows it to adapt to changes in the environment during the course of an infection [2]. Other pathogenic mycobacteria cause diseases such as  [9]. leprosy, evoked by M. leprae, or buruli ulcer, due to infection by M. ulcerans [3,4]. In addition, numerous nonpathogenic and facultative pathogenic mycobacteria exist.
Although several diagnostic measures have been developed for TB diagnosis [5,6], most of these techniques are expensive, e.g., immunological tests using antigens, DNA analysis, or specific culturing conditions, especially for developing countries with the highest burden of TB [7][8][9]. Thus, the analyses of speciesspecific volatiles obtained from breath samples of potentially infected individuals, or from the bacteria themselves, have recently been proposed for rapid diagnosis of TB [8][9][10][11]. The feasibility of these methods is supported by reports that trained Cricetomys rats can distinguish between sputum samples from TB-infected and noninfected persons [12,13].
While the first studies based on electronic sensors or fuzzylogic experiments gave promising results [8,10], recent studies have focused on the characteristic compounds released by pathogenic M. bovis and M. tuberculosis strains [9,11]. Compounds from these species were also expected to occur in breath samples from TB patients [8,10]. In these studies, volatiles were collected by using solid-phase micro-extraction (SPME). Methyl phenylacetate (1), methyl p-anisate (2), methyl nicotinate (3), and o-phenylanisole (4) were emitted by pathogenic M. bovis and M. tuberculosis (Figure 1), while nontuberculous mycobacteria did not produce these four compounds [9]. Due to a central role in mycobacterial metabolism [14], nicotinic acid, closely related to (3), was used in these studies as a reference compound from the breath of TB-positive patients [11]. Infected individuals also released nicotinic acid (excluding smoking patients in this study), proving the utility of this approach [11].
Nevertheless, a broad analysis of the volatiles produced by M. tuberculosis, nontuberculous mycobacteria and mycobacteriarelated species has not been performed. This knowledge is of utmost importance in the identification of potential compounds for use in pattern-recognition methods, such as electronic noses and trained Cricetomys rats. We report here on the identification of volatiles produced by different strains of M. tuberculosis, as well as of M. smegmatis, M. aurum, M. neoaurum, M. aichiense, M. scrofulaceum, M. avium spp. avium, M. vaccae, Nocardia africana, and N. asteroides, using closed-loop stripping analysis (CLSA) for the collection of volatile compounds [15,16]. This method allows sampling for longer periods than SPME, and usually results in a lower detection limit. Compounds not detectable by SPME can thus be detected. SPME preferentially favors less volatile compounds [17]. Furthermore, SPME may have a discriminative effect, and our experience has shown that minor components are often not detected when large amounts of a major component are present.
The investigated mycobacteria were in different stages of their lifecycles and grown on various different media. The volatiles released under these conditions were identified and their formation under different conditions is discussed.

Results and Discussion
Volatiles released from different mycobacteria and Nocardia spp. grown on solid or in liquid media were collected by using CLSA as described previously [15,16]. The headspace extracts were analyzed by gas-chromatography-mass-spectrometry (GC-MS). Compound identification was performed by comparison of mass spectra and gas-chromatographic retention indices with those of authentic reference compounds and massspectra libraries [18].

Analysis of bacteria grown on a solid medium
Different strains of M. tuberculosis grown on a 7H11 solid medium were analyzed after different incubation periods (days). The detection of bacteria-specific compounds was assured by the analysis of headspace samples from a sterile medium incubated in parallel with bacterial strains. Although many compounds were released from the medium, bacteria-specific compounds were distinguishable. A typical gas chromatogram is shown in Figure 2, revealing that many compounds are formed from the nutrient medium that is necessary to grow the bacteria. The identified compounds are listed in Table 1.
As shown in Table 1, early cultures produced few compounds; only those older than 18 days produced a bouquet of compounds. This is probably due to the slow growth of many mycobacteria [7,20]. Increased production of volatile compounds occurred in the stationary phase (i.e., starting after three weeks of growth), a phenomenon that corroborates the recent findings that M. tuberculosis culture is detected by trained rats mostly at this growth stage [20]. A full description of the growth stages used in this report has recently been published [20]. avium, and M. vaccae revealed that these diverse nontuberculous mycobacteria did not produce as many different compounds as M. tuberculosis (Table 2), and what they did produced was often in lower quantities. The exceptions are M. scrofulaceum and M. avium ssp. avium, which cause cervical lymphadenitis in children and avian TB, as well as opportunistic infections in immunocompromised humans [21,22].
Although certain volatile, aromatic compounds, such as 1, 6, 8, and 9, were produced in some cases, many compounds from extracts of M. tuberculosis were absent in the other mycobacteria. The low emission rate of volatile compounds was not a result of the early stage at which the mycobacteria were analyzed, since all strains were fast-growing [20,21,[23][24][25][26][27][28].
While the mycobacteria grown on the 7H11 solid medium emitted a wide variety of volatiles, Nocardia asteroides produced, even in different ages, only 4-pentanolide (10) ( Table 3). Additionally, N. africana also released only one compound, the sesquiterpene aciphyllene (19), which is a known volatile from several plants [29], and the endophytic fungus Muscodor albus [30]. The structure of 19 has recently been revised [29]. This compound may act as marker for norcardiae,    as it was produced as a single compound by Nocardia spp. grown on a 7H11 medium and has so far not been reported for other bacteria [19].

Analysis of bacteria grown in diverse liquid media
In these experiments, two different media were used for culturing bacteria. Nocardia spp. were only cultured on the 7H9 broth medium, and all mycobacteria were grown on the 7H9 broth and the Sauton liquid media. The results of the analyses of different strains of M. tuberculosis grown on the 7H9 broth medium are shown in Figure 4 and Table 4.
A 33-day-old culture of M. tuberculosis strain 2, and a 6-dayold culture of strain 5 produced the largest variety of compounds. Again, aromatic compounds including the previously observed volatiles 1, 5, 7, and 8, as well as the fatty acid derivative 10, were present in the headspace extracts.   Frequently, mycobacteria are observed to grow faster in liquid medium than on solid medium [31,32]. Consistent with this, M. tuberculosis strain 5 produced the most volatiles after six days of growth, unlike cultures of the same age grown on solid medium. Heavy inoculation of a fresh medium with a liquid culture of M. tuberculosis at the exponential growth phase may also generate a high yield of volatiles. Several volatiles released in the solid-medium experiments were not observed in liquid culture, and vice versa. On a solid medium, compounds 14-18 were formed but were not emitted in liquid cultures. In contrast, liquid cultures of M. tuberculosis released 1-hexanol (20), phenol (21), camphor (22), benzyl acetate (23), methyl dimethylbenzoate (24), and a methyl butenolide, all of which were absent in strains cultured on a solid medium.
The headspace extracts of the nontuberculous mycobacteria cultured in the 7H9 broth medium were investigated and the results ( The analyses of the Nocardia strains revealed interesting differences between the two media types. Although aciphyllene (19) was again produced by N. africana, it was also present in the bouquet of an early culture of N. asteroides (Table 6). Additionally, both bacteria produced several unknown diterpenoids. Such relatively large compounds have rarely been reported as bacterial volatiles [33,34].
In additional experiments, a Sauton liquid medium was used to culture M. tuberculosis ( Table 7). The two strains of M. tuberculosis produced fewer volatiles compared to cultures on   avium) showed that fewer compounds were produced by these species in this medium ( Table 8).
As observed for the 7H9 medium, only M. avium ssp. avium emitted a diverse array of volatiles, while the other mycobac-teria produced a few compounds only. The results of different analyses showed that M. tuberculosis produces more compounds than the previously described volatiles 1-4, especially on the solid medium.
Apart from the production of specific compounds such as 3 and 4, M. tuberculosis was characterized by a more pronounced production of volatiles compared to other mycobacteria and associated Nocardia bacteria. While many of these compounds, such as 2-phenylethanol, 1, or 8, are commonly found in some, but not all, bacteria [19], the individual components contribute to a specific bouquet of volatiles. This bouquet probably enables olfactory detection of M. tuberculosis by trained Cricetomys rats [13], or could enable potential detection by electronic noses. Only M. scrofulaceum and M. avium ssp. avium were found to produce a nutrient mixture of volatiles while other strains did not. The volatiles identified belonged predominantly to the biosynthetic class of aromatic compounds, while metabolites of the fatty-acid-biosynthesis pathway were also present.
Since M. tuberculosis grown in the Sauton liquid medium produced only a few volatiles after more than 25 days, these experimental conditions are suboptimal. M. tuberculosis cultivated in the 7H9 broth medium emitted after a shorter incubation time more volatiles in comparison to cultures grown in solid medium. However, the number of volatiles produced compared to nontuberculous mycobacteria was much lower than for mycobacteria grown on solid medium. The compounds emitted by M. tuberculosis grown on the 7H9 broth medium are known, except for 3, but are also released from other bacteria, thus diminishing their diagnostic potential [19,35,36]. Compound 19 may play a role as a marker for different Nocardia strains, depending on growth conditions. Further studies should focus on enhancing the production of volatiles in liquid medium, which support the rapid growth of bacteria. Currently the identified volatile compounds produced by M. tuberculosis are tested for tuberculosis detection by using Cricetomys rats.

Conclusion
In conclusion, profound qualitative and quantitative differences (number of compounds as well as probably higher emission rates) in the bouquet of volatiles from M. tuberculosis, nontuberculous mycobacteria and other bacteria grown on different media were found. M. tuberculosis produces a distinctive bouquet of compounds, consisting of compounds known to be produced by other bacteria, but also including relatively specific compounds such as methyl nicotinate (3). Variations occurred within individual analyses and also between different media. Therefore, it seems unlikely that GC-MS analyses of individual cultures can be used as a diagnostic tool. Nevertheless, a system able to detect mixtures within a given compositional tolerance seems more promising. In this sense, trained Cricetomys rats seem to be well suited for detection [20]. In a more technical variant, experiments with electronic noses, able to discriminate between different odor profiles, seem to be more promising for the detection of Mtb-specific odor profiles and, thus, could be potentially used for TB diagnosis. For bacteria grown in the 7H9 liquid medium, the inoculum (500 µL) was aseptically transferred from actively growing cultures and inoculated into fresh medium (15-20 mL).

Experimental
Cultures were incubated at 28-37 °C under aerobic conditions.
Bacterial cultures for inoculation into Sauton medium (without glycerol) were washed in order to remove traces of ingredients from the stock-culture medium (7H9). Washing was done three times by mixing the culture (3 mL) with phosphate-buffered saline (PBS; 10 mL) and centrifuging at 4000 rpm for 10 min. Supernatants were decanted and PBS (10 mL) was added to the pellet, which was dissolved by careful pipetting. The suspension was centrifuged again at 4000 rpm for 10 min. The final pellet was suspended in sterile PBS (4 mL), mixed thoroughly and 500 µL of the suspension was subcultured into fresh Sauton medium (30 mL). The Sauton medium consisted of the following ingredients in 1000 mL of distilled water: asparagine (4 g), MgSO 4 (0.5 g), K 2 PO 4 (0.5 g), citric acid (1.83 g), ferric ammonium citrate (0.05 g), D-(+)-glucose monohydrate (4.82 g), and sodium pyruvate (4.82 g). The pH was adjusted to 6.8 and the medium was filter-sterilized by using a 0.22 µm membrane filter (Millipore Corp., USA). Cultures were incubated at 37 °C aerobically.

Sampling of volatiles
Volatile organic compounds emitted by cell cultures of the different mycobacteria were collected by using the CLSA technique [14,15]. The volatiles were adsorbed on charcoal (Chromtech; Precision Charcoal Filter, 5 mg) for 24 h, and then eluted with 30 µL of CH 2 Cl 2 . The obtained extracts were immediately analyzed by GC-MS, and stored at -30 °C. Retention indices (I) were determined from a homologous series of n-alkanes (C8-C35) [19]. Identification of compounds was performed by comparison of mass spectra to the Wiley-6 Library, NIST 07, and the Essential Oils Library (Massfinder) and gas chromatographic retention indices, as well as by comparison with synthetic samples. Details can be found in the supporting information. The relative emission of volatiles was roughly estimated from peak areas, although overlapping peaks from the medium and the known difficulty in using CLSA for quantification allowed for only a rough approximation.

Supporting Information
Supporting Information File 1 Identification of compounds.